VIII.
RADIO ASTRONOMY
Prof. A. H. Barrett
Prof. J. W. Graham
A. T. Anderson III
J.
V.
J.
V.
W.
J.
C.
K.
Barrett
Bates
Blinn III
Chung
J. R. Cummings
D. H. Staelin
S. Weinreb
AN ATTEMPT TO MEASURE ZEEMAN SPLITTING OF THE GALACTIC
A.
21-cm HYDROGEN LINE
1.
Introduction
The existence of a galactic magnetic field is presupposed by theories describing the
polarization of starlight, the emission of galactic radio noise, and the behavior of cosmic rays. The magnitude of the field cannot be measured directly from these effects,
although estimates of from 10-5 gauss to 10-6gauss are common.
cussed and referenced by Shklovsky,
l
and also in two papers,
These facts are dis-
3 that describe previous
attempts to measure the field by measuring the Zeeman splitting of the 21-cm hydrogen
line.
The experiment was first suggested by Bolton and Wild. 4
A brief description of the Zeeman effect on hydrogen line radiation may be given as
follows.
The line-of-sight component of the galactic magnetic field causes a splitting of
the radiation into left-hand and right-hand circularly polarized waves with a difference
-66
in frequency of 2. 8 cps per 10 -
gauss.
The sharpest hydrogen lines found in absorp-
tion of the strong discrete sources are ~10-20 kc wide,
3-30 cps is quite difficult to measure.
and thus the expected splitting of
The procedure consists of alternately measuring
the hydrogen profile with feeds that are receptive to circular polarization of opposite
sense and recording the difference profile.
This difference profile, AT(f),
is related
to the observed (either polarization) profile, T(f), by the following relation which holds
if the frequency splitting, Af, is small compared with the width of the observed line:
AT(f) = Af - T'(f) = 2.8 X 10 6 H - T'(f),
where H is the magnetic field, and the prime denotes derivative with respect to frequency.
sources,
2.
The observed profile and the expected difference profile for the two observed
Cas A and Taurus A, are shown in Figs. VIII-1 and VIII-2.
Experimental Procedure
Observations of Cas A and Taurus A were performed with the use of the 85' Howard
Tatel Radio Telescope at the National Radio Astronomy Observatory,
Green Bank,
West Virginia, in conjunction with a dual circular polarization feed built by Jasik Laboratories.
Switching between polarizations was done at a
1/14 cps rate by using a
*This work was supported in part by the National Science Foundation (Grant G-13904);
and in part by N. A. S. A. Contract Nasr-101.
CAS A
OBSERVED PROFILE - CAS A
ZEEMAN PROFILE
0.20
F
FOR 5 x 10- 6 GAUSS
0 .260
-0.20
DATA ADATA A
0.20
DEVIATION = 0.0 730
DATA B
0.20
THEORETICAL RMS
THEORETICAL RMS
DEVIATION = 0.0530
10 KC
-20 KC
FREQUENCY -.
Fig. VIII- 1.
Results of attempt to measure Zeeman splitting of 21-cm galactic
hydrogen line observed in absorption of Cas A radio source.
TAURUS A
OBSERVED PROFILE - TAURUS A
600
0.20
-0.2
ZEEMAN PROFILE
°
0.20
THEORETICAL RMS
DEVIATION = 0.0440
DATA A
-0. 20
THEORETICAL RMS
DEVIATION = 0.080
0.20
DATA B
70 KC
-60 KC
-50 KC
-Q KCC
\-30KC
FREQUENCY
Fig. VIII-2.
Results of attempt to measure Zeeman splitting of 21-cm galactic
hydrogen line observed in absorption of Taurus A radio source.
(VIII.
RADIO ASTRONOMY)
mechanical coaxial switch.
Double stub tuners were inserted between each arm of the
switch and the feed, and were adjusted so that the standing-wave ratio of each polarization feed was less than 1.01 when measured through the switch.
An electron-beam, traveling-wave, parametric amplifier was used for all observations except for those of Taurus, Data B, for which a crystal mixer was used. The
single-channel noise temperature, including switch and stub losses, was 6000 for the
parametric amplifier, and 10800 for the crystal mixer.
Through conventional means a 50-kc portion of the band is filtered out and shifted
down to low frequencies at which its autocorrelation function is found by a digital correlator. 5 A computer Fourier transform of the autocorrelation function then gives the
power spectrum.
This system is analogous to a multichannel radiometer having 7.5-kc
wide filters spaced over the 50-kc band.
The observational procedure consisted, first, of making a 10-minute observation of
the absorption profile, with the use of the right-hand circular polarization feed. Typical
results are the top curves in Figs. VIII-1 and VIII-2. A loss in sensitivity or a frequency
error could be found in this manner. Next, after this run, a 5-hour run for measuring
the difference profile was made. The local oscillator was reset every half hour to correct for Doppler shift resulting from the earth's motion. One or two 5-hour runs were
made on Cas A and Taurus A each day for approximately 35 days.
3.
Results
A difficulty that arises in performing the Zeeman experiment is due to small differences in gain between the right-hand and left-hand circular polarization feeds. If the
gain of one feed is 1 + a times the gain of the other feed, then the difference profile,
AT(f), will contain a term a - T(f); the absorption profile will appear in the data. Since
the absorption profile is known, it may be removed from the difference profile. However,
care must be taken that the correction is not shifted in frequency, for this would cause
a false Zeeman effect. It is easily seen that if an unbalance signal, a - T(f), is corrected
by subtraction of a profile, a - T(f+6f), then the result is the same as that which would
be found for a Zeeman splitting of value a - 6f.
Approximately 20 per cent of the data that were taken showed values of a less than
0. 003 (Data A); 40 per cent had a between 0. 003 and 0. 03 (Data B); and the rest had
a greater than 0. 03 and were not used. The variations in a are probably due to temperature effects on the feed, stubs, and switch, and to tracking errors and flexure of the
telescope.
The largest error in the local oscillators was less than 100 cps; thus even
with a = 0. 03, the largest spurious Zeeman effect would be 3 cps, or ~10 - 6 gauss.
The average of the Data A and Data B runs for Cas A and Taurus A are plotted
in the figures. The Data A runs require no correction for feed unbalance, while the
Table VIII-1.
Theoretical
RMS
Fluctuation
.0730
Cas
Actual
RMS
Fluctuationt
.099 0
Zeeman results.
Total
Noise
Temperature
T
Profile
Depth
Observed
Time
Treatment of Data
10600
1800
1.2 x 105
No correction to data
10600
1800
2. 1 X 105
Absorption profile of
1.370 peak subtracted
from raw data
T
(sec)
Data A
1.4
1.9
Cas
Data B
. 0530
1.0
. 12
2.3
Taurus
Data A
.0440
2.2
.040
2. O0
8500
600
2.2 X 105
Small slope correction
of 0. 120/10 kc
Taurus
Data B
.080
4.0
.050
2. 5
13300
600
1.6 X 105
Absorption profile of
1.150 peak subtracted
from raw data
AT = 2. 1 T/
B,
Observation Time.
a peak signal = AT.)
°
where T = Antenna + Receiver Noise Temperature,
B = 7. 5 X 103,
and T is the
(The quantity below AT is the magnetic field in units of 10 6 gauss which will give
tAT' = rms value of data, with the mean taken over frequency.
(VIII.
RADIO ASTRONOMY)
Data B runs have been corrected as indicated in Table VIII-1.
Our conclusion is that
the line-of-sight component of the magnetic field is less than 3 X 10 and less than 5 X 10
-6
6
gauss in Cas A,
gauss in Taurus A.
This work was supported in part by the National Radio Astronomy Observatory.
S. Weinreb
References
1. I. S. Shklovsky, Cosmic Radio Waves (Harvard University Press, Cambridge,
Mass., 1960), pp. 182-190.
2. J. A. Galt, C. H. Slater, and W. L. H. Shuter, An attempt to detect the galactic
magnetic field using Zeeman splitting of the hydrogen line, Mon. Not. Royal Astro.
Soc. 120, 187-192 (1960).
3. R. D. Davies, C. H. Slater, W. L. H. Shuter, and P. A. T. Wild, A new limit
to the galactic magnetic field set by measurements of the Zeeman splitting of the hydrogen line, Nature 187, 1088-1089 (September 24, 1960).
4. J. G. Bolton and J. P. Wild, On the possibility of measuring interstellar magnetic fields by 21-cm Zeeman splitting, Astrophys. J. 125, 296 (1957).
5.
S. Weinreb, Digital radiometer,
Proc. IRE 49,
1099 (1961).